{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:YDNCYSZEWHLLSNVSI3IPXGBJQP","short_pith_number":"pith:YDNCYSZE","schema_version":"1.0","canonical_sha256":"c0da2c4b24b1d6b936b246d0fb982983f60cbbb21c8b0110cc64b437dd0c5c04","source":{"kind":"arxiv","id":"2507.18809","version":2},"attestation_state":"computed","paper":{"title":"Test-time Offline Reinforcement Learning on Goal-related Experience","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Andreas Krause, Georg Martius, Jonas H\\\"ubotter, Marco Bagatella, Mert Albaba","submitted_at":"2025-07-24T21:11:39Z","abstract_excerpt":"Foundation models compress a large amount of information in a single, large neural network, which can then be queried for individual tasks. There are strong parallels between this widespread framework and offline goal-conditioned reinforcement learning algorithms: a universal value function is trained on a large number of goals, and the policy is evaluated on a single goal in each test episode. Extensive research in foundation models has shown that performance can be substantially improved through test-time training, specializing the model to the current goal. We find similarly that test-time "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2507.18809","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-07-24T21:11:39Z","cross_cats_sorted":[],"title_canon_sha256":"ea51ed3063d1bf0826633598e375d49a5aff20700692c044e33f8ca422c8fb6a","abstract_canon_sha256":"31b05c4e4ad2ad4ef6c868fd12e1f1d3006b033a614144ba34d0928581d3f942"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:09:34.981693Z","signature_b64":"mbrdAEjyLg5H387v14IpPYMiN5ysFJZ9mMAA8xExGkNWfH7vqrm5pOTiEJnx7iS6k96tE2D5LmN3fOO8JnljBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c0da2c4b24b1d6b936b246d0fb982983f60cbbb21c8b0110cc64b437dd0c5c04","last_reissued_at":"2026-05-18T03:09:34.981253Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:09:34.981253Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Test-time Offline Reinforcement Learning on Goal-related Experience","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Andreas Krause, Georg Martius, Jonas H\\\"ubotter, Marco Bagatella, Mert Albaba","submitted_at":"2025-07-24T21:11:39Z","abstract_excerpt":"Foundation models compress a large amount of information in a single, large neural network, which can then be queried for individual tasks. There are strong parallels between this widespread framework and offline goal-conditioned reinforcement learning algorithms: a universal value function is trained on a large number of goals, and the policy is evaluated on a single goal in each test episode. Extensive research in foundation models has shown that performance can be substantially improved through test-time training, specializing the model to the current goal. We find similarly that test-time "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.18809","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2507.18809","created_at":"2026-05-18T03:09:34.981321+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.18809v2","created_at":"2026-05-18T03:09:34.981321+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.18809","created_at":"2026-05-18T03:09:34.981321+00:00"},{"alias_kind":"pith_short_12","alias_value":"YDNCYSZEWHLL","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"YDNCYSZEWHLLSNVS","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"YDNCYSZE","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2510.10181","citing_title":"Dejavu: Towards Experience Feedback Learning for Embodied Intelligence","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2601.16175","citing_title":"Learning to Discover at Test Time","ref_index":4,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YDNCYSZEWHLLSNVSI3IPXGBJQP","json":"https://pith.science/pith/YDNCYSZEWHLLSNVSI3IPXGBJQP.json","graph_json":"https://pith.science/api/pith-number/YDNCYSZEWHLLSNVSI3IPXGBJQP/graph.json","events_json":"https://pith.science/api/pith-number/YDNCYSZEWHLLSNVSI3IPXGBJQP/events.json","paper":"https://pith.science/paper/YDNCYSZE"},"agent_actions":{"view_html":"https://pith.science/pith/YDNCYSZEWHLLSNVSI3IPXGBJQP","download_json":"https://pith.science/pith/YDNCYSZEWHLLSNVSI3IPXGBJQP.json","view_paper":"https://pith.science/paper/YDNCYSZE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.18809&json=true","fetch_graph":"https://pith.science/api/pith-number/YDNCYSZEWHLLSNVSI3IPXGBJQP/graph.json","fetch_events":"https://pith.science/api/pith-number/YDNCYSZEWHLLSNVSI3IPXGBJQP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YDNCYSZEWHLLSNVSI3IPXGBJQP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YDNCYSZEWHLLSNVSI3IPXGBJQP/action/storage_attestation","attest_author":"https://pith.science/pith/YDNCYSZEWHLLSNVSI3IPXGBJQP/action/author_attestation","sign_citation":"https://pith.science/pith/YDNCYSZEWHLLSNVSI3IPXGBJQP/action/citation_signature","submit_replication":"https://pith.science/pith/YDNCYSZEWHLLSNVSI3IPXGBJQP/action/replication_record"}},"created_at":"2026-05-18T03:09:34.981321+00:00","updated_at":"2026-05-18T03:09:34.981321+00:00"}